Add waterfall plot for local explanations

[lucawsanchez]

Hi all,

Love the library, thank you!

I was wondering if it might be possible to add an option for a waterfall plot for local explanations in explain_local, which could be called through interpret.show? They are a great way of visualising how the model goes from intercept to prediction. It is currently possible to extract all the necessary data and build your own, and maybe that's how it should remain to avoid over complicating the API, but I thought it would be an interesting enhancement to consider.

For example, this is done in the shap package and the ExplainerDashboard package (where it's called a contributions plot).

All the best,

Luca

[paulbkoch]

Hi @lucawsanchez -- Glad to hear you're enjoying the library. Having an option to do Waterfall plots would be great. I've mostly been focusing on the lower levels of EBMs lately, so if this were to be added in the near term it would have to come from a PR.